V.A.L.L.Y.: An Ultra-Fast Computational Framework Integrating Elastic Network Models and Informed Heuristics for Accelerating Drug Discovery

The high computational cost of molecular simulation is a critical bottleneck in drug discovery. To address this, we developed V.A.L.L.Y. (Vibrational Analysis for Ligand Likelihood Yielding), a computational framework for speed and accessibility, implemented as the open-source tool VALLY-Scan. Our method performs an ultra-fast Vibrational Analysis of protein structures using Anisotropic Network Models (ANM) to capture essential collective dynamics. This analysis, completed in seconds on standard hardware, feeds a novel "Informed Heuristic Predictor" that integrates expert knowledge from scientific literature to evaluate and rank potential therapeutic sites. We validated this framework on two high-priority viral targets: the Dengue Virus protease (PDB: 2FOM) and the SARS-CoV-2 main protease (PDB: 6LU7). Our results identify key residues with high dynamic impact, distinct from the catalytic sites, suggesting potential allosteric sites for drug design. This work demonstrates a powerful, first-tier methodology for rapidly generating therapeutic hypotheses, thereby democratizing computational biophysics.